def __init__(self, username="", apikey="", version="latest", level="lv2"):
'''
username = "" # 你註冊時的 email。若留空，則會使用每小時更新 2000 字的公用帳號。
apikey = "" # 您完成付費後取得的 apikey 值。若留空，則會使用每小時更新 2000 字的公用帳號。
'''
try:
with open("./account.info", "r") as f:
userDICT = json.loads(f.read())
self.username = userDICT["email"]
self.apikey = userDICT["apikey"]
except:
self.username = username
self.apikey = apikey
self.url = "https://api.droidtown.co"
self.version = version
self.level = level
self.userDefinedDictFILE = None
self.openDataPlaceAccessBOOL = False
self.chemicalBOOL = True
self.fileSizeLimit = 1024 * 1024 * 10 # 10 MB
# Toolkit
self.analyse = AnalyseManager()
self.localRE = TaiwanAddressAnalizer(locale="TW")
def __init__(self, url="http://127.0.0.1", port="8964", bulkSize=20, userDefinedDictFILE=None):
self.port = port
if "http" not in url:
self.url = "http://{}:{}".format(url, port)
else:
self.url = "{}:{}".format(url, port)
self.bulkSize = bulkSize
self.processes = cpu_count()
if self.processes > 8:
self.processes = 8
self.userDefinedDictFILE = None
self.openDataPlaceAccessBOOL=False
self.fileSizeLimit = 1024 * 1024 * 10 # 10 MB
self.userDefinedDICT = {}
if userDefinedDictFILE:
try:
if os.path.getsize(userDefinedDictFILE) <= self.fileSizeLimit:
userDefinedFile = json.load(open(userDefinedDictFILE, "r", encoding="utf8"))
if type(userDefinedFile) == dict:
self.userDefinedDICT = userDefinedFile
else:
print("User Defined File must be dict type.")
return {"status": False, "msg": "UserDefinedDICT Parsing ERROR. Please check your the format and encoding."}
else:
print("Maximum file size limit is 10 MB.")
except Exception as e:
print("User Defined File Loading Error.")
print(str(e))
return {"status": False, "msg": "UserDefinedDICT Parsing ERROR. Please check your the format and encoding."}
# Toolkit
self.analyse = AnalyseManager()
self.localRE = TaiwanAddressAnalizer(locale="TW")
def __init__(self, username="", apikey="", version="latest", level="lv2"):
'''
username = "" # 你註冊時的 email。若留空，則會使用每日 1 萬字的公用帳號。
apikey = "" # 您完成付費後取得的 apikey 值。若留空，則會使用每日 1 萬字的公用帳號。
'''
try:
with open("./account.info", "r") as f:
userDICT = json.loads(f.read())
self.username = userDICT["email"]
self.apikey = userDICT["apikey"]
except:
self.username = username
self.apikey = apikey
self.url = "https://api.droidtown.co"
self.version = version
self.level = level
self.userDefinedDictFILE = None
self.openDataPlaceAccessBOOL = False
self.fileSizeLimit = 1024 * 1024 * 10 # 10 MB
self.verbPPat = re.compile("(?<=<VerbP>)[^<]*?(?=.</VerbP>)")
self.verbPat = re.compile("(?<=<ACTION_verb>)[^<]*?(?=</ACTION_verb>)")
self.nounPat = re.compile(
"(?<=<ENTITY_nounHead>)[^<]*?(?=</ENTITY_nounHead>)|(?<=<ENTITY_nouny>)[^<]*?(?=</ENTITY_nouny>)|(?<=<ENTITY_noun>)[^<]*?(?=</ENTITY_noun>)|(?<=<ENTITY_oov>)[^<]*?(?=</ENTITY_oov>)"
)
self.modifierPat = re.compile("(?<=<MODIFIER>)[^<]*?(?=</MODIFIER>)")
self.funcPat = re.compile(
"(?<=<AUX>)[^<]*?(?=</AUX>)|(?<=<FUNC_in[nt]er>)[^<]*?(?=</FUNC_in[nt]er>)|(?<=<RANGE_locality>)[^<]*?(?=</RANGE_locality>)|(?<=<RANGE_period>)[^<]*?(?=</RANGE_period>)"
)
self.personPat = re.compile(
"(?<=<ENTITY_person>)[^<]*?(?=</ENTITY_person>)")
self.pronounPat = re.compile(
"(?<=<ENTITY_pronoun>)[^<]*?(?=</ENTITY_pronoun>)")
self.locationPat = re.compile("(?<=<LOCATION>)[^<]*?(?=</LOCATION>)")
self.placePat = re.compile(
"(?<=<KNOWLEDGE_place>)[^<]*?(?=</KNOWLEDGE_place>)")
self.timePat = re.compile(
"(?<=<TIME_decade>)[^<]*?(?=</TIME_decade>)|(?<=<TIME_year>)[^<]*?(?=</TIME_year>)|(?<=<TIME_season>)[^<]*?(?=</TIME_season>)|(?<=<TIME_month>)[^<]*?(?=</TIME_month>)|(?<=<TIME_week>)[^<]*?(?=</TIME_week>)|(?<=<TIME_day>)[^<]*?(?=</TIME_day>)|(?<=<TIME_justtime>)[^<]*?(?=</TIME_justtime>)"
)
self.eventPat = re.compile(
"<ACTION_lightVerb>[^<]</ACTION_lightVerb>((?<!</LOCATION>)<ENTITY_nouny>[^<]</ENTITY_nouny>)?((<ACTION_verb>[^<]*?</ACTION_verb>)|(<ENTITY_nouny?>[^<]*?</ENTITY_nouny?>$))|<ACTION_verb>.?[有現到見道]</ACTION_verb>(<ENTITY_nouny?>[^<]*?</ENTITY_nouny?>|<ENTITY_person>[^<]*?</ENTITY_person>)$|(?<=[有現到見道]</ACTION_verb>)((?<!</LOCATION>)<ENTITY_nouny?>[^<]*?</ENTITY_nouny?>)?<ACTION_verb>[^<有現到見道]{1,2}</ACTION_verb>$|((?<!</LOCATION>)<ENTITY_nouny?>[^<]*?</ENTITY_nouny?>)?<ACTION_verb>[^<有現到見道]{1,2}</ACTION_verb>(?!<ACTION)(?!<LOCATION)(?!<KNOWLEDGE)(?!<MODIFIER>)(?!<ENTITY_classifier)(?!<ENTITY_pronoun>)(<ENTITY_nouny?>[^<]*?</ENTITY_nouny?>|<ENTITY_person>[^<]*?</ENTITY_person>)?|<ACTION_lightVerb>.</ACTION_lightVerb><VerbP>[^<]*?</VerbP>|<ACTION_verb>[^<]*?</ACTION_verb>($|(?=<ACTION_verb>))"
)
self.addTWPat = re.compile(
"(?<=<KNOWLEDGE_addTW>)[^<]*?(?=</KNOWLEDGE_addTW>)")
self.stripPat = re.compile("(?<=>).*?(?=<)")
self.clausePat = re.compile("\<CLAUSE_.*?Q\>")
self.contentPat = re.compile("|".join([
self.verbPat.pattern, self.nounPat.pattern,
self.modifierPat.pattern, self.verbPPat.pattern
]))
# Toolkit
self.analyse = AnalyseManager()
self.localRE = TaiwanAddressAnalizer(locale="TW")
try:
self.graphQL = GraphQL()
except:
pass
def __init__(self,
url="ws://127.0.0.1",
port="8964",
bulkSize=20,
userDefinedDictFILE=None):
self.port = port
if url.startswith("ws"):
self.ws_url = "{}:{}/Articut/WebSocket".format(url, port)
self.url = "{}:{}".format(url.replace("ws", "http"), port)
elif url.startswith("http"):
self.ws_url = "{}:{}/Articut/WebSocket".format(
url.replace("http", "ws"), port)
self.url = "{}:{}".format(url, port)
else:
self.ws_url = "ws://{}:{}/Articut/WebSocket".format(url, port)
self.url = "http://{}:{}".format(url, port)
#enableTrace(True)
self.ws = create_connection("{}/API/".format(self.ws_url))
self.ws_bulk = create_connection("{}/BulkAPI/".format(self.ws_url))
self.bulkSize = bulkSize
self.userDefinedDictFILE = None
self.openDataPlaceAccessBOOL = False
self.fileSizeLimit = 1024 * 1024 * 10 # 10 MB
self.userDefinedDICT = {}
if userDefinedDictFILE:
try:
if os.path.getsize(userDefinedDictFILE) <= self.fileSizeLimit:
userDefinedFile = json.load(
open(userDefinedDictFILE, "r", encoding="utf8"))
if type(userDefinedFile) == dict:
self.userDefinedDICT = userDefinedFile
else:
print("User Defined File must be dict type.")
return {
"status":
False,
"msg":
"UserDefinedDICT Parsing ERROR. Please check your the format and encoding."
}
else:
print("Maximum file size limit is 10 MB.")
except Exception as e:
print("User Defined File Loading Error.")
print(str(e))
return {
"status":
False,
"msg":
"UserDefinedDICT Parsing ERROR. Please check your the format and encoding."
}
# Toolkit
self.analyse = AnalyseManager()
self.localRE = TaiwanAddressAnalizer(locale="TW")
self.LawsToolkit = LawsToolkit()
self.NER = GenericNER()
self.POS = ArticutPOS()
def __init__(self, username="", apikey="", version="latest", level="lv2"):
'''
username = "" # 你註冊時的 email。若留空，則會使用每小時更新 2000 字的公用帳號。
apikey = "" # 您完成付費後取得的 apikey 值。若留空，則會使用每小時更新 2000 字的公用帳號。
'''
try:
with open("./account.info", "r") as f:
userDICT = json.loads(f.read())
self.username = userDICT["email"]
self.apikey = userDICT["apikey"]
except:
self.username = username
self.apikey = apikey
self.url = "https://api.droidtown.co"
self.version = version
self.level = level
self.userDefinedDictFILE = None
self.openDataPlaceAccessBOOL = False
self.fileSizeLimit = 1024 * 1024 * 10 # 10 MB
# Regex Pattern
self.verbPPat = re.compile("(?<=<VerbP>)[^<]*?(?=.</VerbP>)")
self.verbPat = re.compile("(?<=<ACTION_verb>)[^<]*?(?=</ACTION_verb>)")
self.nounPat = re.compile(
"(?<=<ENTITY_nounHead>)[^<]*?(?=</ENTITY_nounHead>)|(?<=<ENTITY_nouny>)[^<]*?(?=</ENTITY_nouny>)|(?<=<ENTITY_noun>)[^<]*?(?=</ENTITY_noun>)|(?<=<ENTITY_oov>)[^<]*?(?=</ENTITY_oov>)"
)
self.modifierPat = re.compile("(?<=<MODIFIER>)[^<]*?(?=</MODIFIER>)")
self.modifierPPat = re.compile(
"(?<=<DegreeP>)[^<]*?(?=</DegreeP>)|(?<=<ModifierP>)[^<]*?(?=</ModifierP>)"
)
self.funcPat = re.compile(
"(?<=<AUX>)[^<]*?(?=</AUX>)|(?<=<FUNC_in[nt]er>)[^<]*?(?=</FUNC_in[nt]er>)|(?<=<RANGE_locality>)[^<]*?(?=</RANGE_locality>)|(?<=<RANGE_period>)[^<]*?(?=</RANGE_period>)"
)
self.personPat = re.compile(
"(?<=<ENTITY_person>)[^<]*?(?=</ENTITY_person>)")
self.pronounPat = re.compile(
"(?<=<ENTITY_pronoun>)[^<]*?(?=</ENTITY_pronoun>)")
self.locationPat = re.compile(
"(?<=<LOCATION>)[^<]*?(?=</LOCATION>)|(?<=<KNOWLEDGE_addTW>)[^<]*?(?=</KNOWLEDGE_addTW>)|(?<=<KNOWLEDGE_routeTW>)[^<]*?(?=</KNOWLEDGE_routeTW>)"
)
self.userDefinedPat = re.compile(
"(?<=<UserDefined>)[^<]*?(?=</UserDefined>)")
self.placePat = re.compile(
"(?<=<KNOWLEDGE_place>)[^<]*?(?=</KNOWLEDGE_place>)")
self.timePat = re.compile(
"(?<=<TIME_decade>)[^<]*?(?=</TIME_decade>)|(?<=<TIME_year>)[^<]*?(?=</TIME_year>)|(?<=<TIME_season>)[^<]*?(?=</TIME_season>)|(?<=<TIME_month>)[^<]*?(?=</TIME_month>)|(?<=<TIME_week>)[^<]*?(?=</TIME_week>)|(?<=<TIME_day>)[^<]*?(?=</TIME_day>)|(?<=<TIME_justtime>)[^<]*?(?=</TIME_justtime>)"
)
self.addTWPat = re.compile(
"(?<=<KNOWLEDGE_addTW>)[^<]*?(?=</KNOWLEDGE_addTW>)")
self.currencyPat = re.compile(
"(?<=<KNOWLEDGE_currency>)[^<]*?(?=</KNOWLEDGE_currency>)")
self.currencyGreedyPat = re.compile(
"(?<=[元金幣圜圓比布索鎊盾銖令朗郎]</ENTITY_noun><ENTITY_num>)[^<]*?(?=</ENTITY_num>)"
)
self.currencyGreedyGapPat = re.compile(
"(?<=^<ENTITY_num>)[^<]*?(?=</ENTITY_num>)")
self.wikiDataPat = re.compile(
"(?<=<KNOWLEDGE_wikiData>)[^<]*?(?=</KNOWLEDGE_wikiData>)")
self.stripPat = re.compile("(?<=>).*?(?=<)")
self.clausePat = re.compile("\<CLAUSE_.*?Q\>")
self.contentPat = re.compile("|".join([
self.verbPPat.pattern, self.verbPat.pattern, self.nounPat.pattern,
self.modifierPat.pattern, self.modifierPPat.pattern,
self.userDefinedPat.pattern
]))
# Toolkit
self.analyse = AnalyseManager()
self.localRE = TaiwanAddressAnalizer(locale="TW")
try:
self.graphQL = GraphQL()
except:
pass